Preliminary native windows support, part 2: standard library and libgerbil #1291
Conversation
👷 Deploy request for elastic-ritchie-8f47f9 pending review.Visit the deploys page to approve it
|
|
@fare first pass again? |
|
General notes
|
| @@ -40,6 +40,13 @@ | |||
|
|
|||
| (begin-ffi (_pipe make_pipe_ptr pipe_ptr_ref) | |||
| (c-declare "#include <unistd.h>") | |||
| (c-declare " | |||
There was a problem hiding this comment.
Shouldn't we be returning -1 for failure?
| #define LOCK_NB 3 | ||
| int flock(int fd, int op) { | ||
| return 0; | ||
| } |
src/std/os/fdio.ss
Outdated
| (c-declare " | ||
| #ifdef _WINDOWS | ||
| int fsync(int fd) { | ||
| return 0; |
src/std/os/fcntl.ss
Outdated
| (c-declare " | ||
| #ifdef _WINDOWS | ||
| int fcntl (int __fd, int __cmd, ...) { | ||
| return 0; |
src/std/os/_socket.scm
Outdated
| @@ -565,6 +582,9 @@ int ffi_socket_sendmsg (int fd, ___SCMOBJ name, ___SCMOBJ io, ___SCMOBJ ctl, int | |||
| msg.msg_flags = 0; | |||
|
|
|||
| return sendmsg (fd, &msg, flags); | |||
| #else | |||
| return 0; | |||
src/std/os/signal.ss
Outdated
| (c-declare " | ||
| #ifdef _WINDOWS | ||
| int kill(int pid, int signo) { | ||
| return 0; |
|
OK for the name, I guess. Beware about the return values, though. I left comments on the patch at ef77b8c |
|
I've corrected the return values (including a few you didn't mention) in the latest commit. Did you find anything missing? |
src/std/build-spec.ss
Outdated
| (gxc: "os/fcntl" ,@(include-gambit-sharp)) | ||
| (gxc: "os/flock" ,@(include-gambit-sharp)) | ||
| (gxc: "os/pipe" ,@(include-gambit-sharp)) | ||
| (gxc: "os/fd" ,@(include-gambit-sharp)) |
There was a problem hiding this comment.
Don't introduce spurious whitespace
There was a problem hiding this comment.
Oops. I was trying to make it prettier and easier to compare them. I'll remove it
| @@ -624,6 +645,9 @@ int ffi_socket_recvmsg (int fd, ___SCMOBJ name, int *rname, ___SCMOBJ io, ___SCM | |||
| *rflags = msg.msg_flags; | |||
|
|
|||
| return r; | |||
| #else | |||
| return -1; | |||
There was a problem hiding this comment.
Maybe we should also errno = EINVAL; or ERROR_NOT_SUPPORTED whenever we return -1; like that so that the error message will be appropriate (unless there's a better E code under Windows)—if it's ERROR_NOT_SUPPORTED we need it defined on the Scheme side, too.
There was a problem hiding this comment.
I'm not a systems programmer (I mostly work on algorithms, graphics, gamedev etc) so finding the "right thing to do" is difficult for me... But since I'll summarize current supported/unsupported things in the upcoming PR (documents and scripts), can we leave these problems for the future?
Much more changes this time :)